A Semantic Web Based Platform for a Medical decision support system

Abstract

Diagnosing Heart diseases is one of the problems that require high level of accurate analysis and prediction. Data sets dealing with the same medical problems such as coronary artery disease (CAD) may show different results when applying the same machine learning technique. Followed by searching for the best combination of classifiers in an ensemble that is generally suitable for all data sets of Heart diseases diagnoses. There exists multiple datasets targeting the same problem. On the other hand, several classifiers are available for the analysis of these data sets. This research aims to analyze the outcome of integrating the results of the common datasets in the same domain and as a second step the classification techniques applied in this domain. As for the first step, the two classifiers applied are decision tree algorithms (fast decision tree and pruned C4.5) the resulted trees are extracted from different data sets and compared. Common features among these data sets are extracted and used in the later analysis for the same disease in any data set. As for the second step, using ensemble methods in decision support systems provide an important help in analyzing this type of diseases. Six classifiers namely used which are (Bayesian Net, Naive Bayes, Multilayer perceptron, Sequential Minimal Optimization Algorithm (SMO), Decision Tree Algorithms (C4.5 and Fast Decision Tree (FDT)) to predict two different heart diseases subsequently. The results show that the classification accuracy of the collected dataset is 78.06% higher than the average of the classification accuracy of all separate datasets, which is 75.48%. On the Other hand, the classification accuracy results for the ensemble reached percentages higher than 90% accuracy. The best ensemble combination appears to be common for both datasets, composed of Bayesian Network, Naive Bayesian, Neural networks , C4.5 and SVM with 94% accuracy .